On Fri, Jan 23, 2015 at 07:59:24PM +0100, John Darrington wrote:
> If you look closely you will see that it is not quite identical.
>
> The SIG / NOSIG flag controls weather significant results should be
> emphasised.
> In the default output mode, PSPP does this by rendering in an italic font.
>
I get the significance levels.
On 22/01/2015 21:51, Alan Mead wrote:
>You're sure "nosig" means to suppress significance levels? Because I
>think in SPSS it's the opposite. Try:
>
>correlations var = v12 to v21/print sig.
>
An identical result !
In both cases I get the significance levels.
On 22/01/2015 21:51, Alan Mead wrote:
You're sure "nosig" means to suppress significance levels? Because I
think in SPSS it's the opposite. Try:
correlations var = v12 to v21/print sig.
This, BTW, makes n
You're sure "nosig" means to suppress significance levels? Because I
think in SPSS it's the opposite. Try:
correlations var = v12 to v21/print sig.
This, BTW, makes no sense to me either.
-Alan
On 1/22/2015 2:48 PM, F. Thomas wrote:
> Hi,
> I try to get a simple co
Hi,
I try to get a simple correlations matrix for some variables labelled
v12 to v21, with no significance levels (if you have 25,000 cases sig
levels become useless).
My command is
correlations var = v12 to v21/print nosig.
And I get significance levels printed.
I think this should be a
Harry Thijssen writes:
> I guess the GUI does the opposite as intended. If you check "Flag
> significant correlations " the command generated says "NOSIG". And
> when unsigned it says "SIG".
Thanks for pointing that out. I've sent out a patch that s
John Darrington writes:
> This is a limitation of the current output
> subsystem. Superscripts, asterisks and the like are not
> currently possible. Hopefully this is going to change soon.
Asterisks should be possible (as long as you don't want them
superscripted).
By the way, I like seeing th
the same. "Seeing
> stars" is a good thing, even if overly-large sample sizes promote Type
> II error.
Actually, I see. PSPP indeed flac significant correlations correctly
(i.e. in italic fonts), but I expected it would flag them with
asterisks. That's why
izes promote Type
> II error.
Actually, I see. PSPP indeed flac significant correlations correctly
(i.e. in italic fonts), but I expected it would flag them with
asterisks. That's why I said it does not flag it at all.
However - I prefer asterisks.
Regards,
Matej
__
; > On Fri, Mar 19, 2010 at 10:38:38PM -0400, Ferguson, Douglas A wrote:
> > I noticed that Bivariate Correlations do not "flag" significant
> correlations, even if you check the checkbox. Is this a bug? I'm running
> the windows version dated March 11.
Perhaps I'm missing something, but I find PSPPIRE calculates Pearson R between
variables very nicely using Analyze-Descriptive Statistics-Crosstabs, and
results agree with SPSS. Or are we talking about a different sort of
correlation?
___
Pspp-users
Somebody sent me an email mentioning, amongst other things, that
CORRELATIONS wasn't implemented, the correlation coefficients used
in CROSSTABS and that T-TEST sometimes returned a p-value of 2.0
Whoever you are, I'm sorry, but I inadvertently deleted your mail
along with all my spam
12 matches
Mail list logo
